SV_MEETING_TITLE

 

 

SPARQL WWW2008 BOF

<scribe> scribe: ericP

ivan: we shoudl not discuss throwing away and starting again
... should be backward compatible

orri: Virtuoso program manager
... clustered query processing
... we would not dream of changinthe SPARLQL spec
... have extended it
... have fedration interests
... can't have SPARQL be less expressive than SQL
... added expression slselects

have aggregation and group by

uri: also OpenLink

ivan; mix of CAD nd translators

scribe: i deal with XML XSL...

chris B: have a php SPARQL impl

scribe: working on database to RDF mapping via SPARQL/SQL rewriter
... currently working on benchmarks

steveH: garlic, built a couple SPARQL enginnes

Tom Heath: interestd from the user perspective

Ivan: built one in pyuthin

ericP:

fabien: INRIA has a SPARQL impl
... in terested in aggregation

simonCh: view mechanism for RDF
... did work on SPARQL semantics
... defined some extensions
... would like extensions for aggregation
... DannyV:

DennyV: philospher
... developer of semantic media ikik

[insert 1]

Andy: SPARQL postponed 12 issues
... UPDATE
... don't want to get tied to a toolkit
... counting
... service description -- lot's of different issues

ivan: whirled peas

andy: all impls have extensions
... what are the few important things that a WG could address
... people ask me what the plan is
... W3C reacts to events like this

ivan: [W3C process overview]
... charter work at end of the year not unrelalistic

ericP: implement, specify, cooperate -- these help you get your interests get met

andy: compatibility is important, but not if it hurts people forever

steveH: i think we bit off more than we could chew
... don't put many featurs in the charter
... we ruled out some extensibility in the syntax
... e.g. no commas between selects
... we were generally anti-comma in the spec
... not fond of default graph/named graph

chris: S tries to meet both groups' needs

orri: we have use cases of querying across UNION in a DB
... expect to extend FROM
... update is fine as andy spec'd it

andy: i documented the SPARUL that i understood the community wanted
... it's not a general purpose graph conversion
... i ran away from transactions
... needs seem to be replicating OLTP

<AndyS> One update proposal: http://jena.hpl.hp.com/~afs/SPARQL-Update.html

andy: what needs do you see?

chrisB: free text search

orri: search engine and SQL world have CONTAINS
... is not a regex

andy: you get ranked matches

Chris: was frustrated by the lack of free text search

ivan: implemented as a boolean
... not a generator

Simon: re page by rank - would be nice to bind cvars by filter functions

ericP: [generator vs. filter]

orri: ... fourth place in the triple pattern

andy: var bindg happens in the pattern
... accessing lists wasn't so useful
... only atomic types frustrates people

ivan: get list querying feedback
... i have seen graph pattern languages

andy: the sparql/inference interaction isn't clear to many people

ivan: user wants to know if an engine can do, say. RDFS inferencing

orri: might have a pragma for enabling
... we do some backward chaining

andy: OWL folks are working on a profile that works on ...

ivan: also one over SQL

Angi: should be in the QL or the system

???: two different URIs for the two different levels of inference?

Danny: what about in the query?

Andrew: folks wanted control when creating the query
... wanted controled inference over ranges of the class hierarchy
... partOf in a cell is ginormous

Ivan: we have use cases for SELECT-COMPLETE-BUT-NOT-SOUND

[migrate to query federation]

[told BNodes]

AndrewN: give the BNode with a CBD

chrisB: radical folks in the community who would like to dispense with BNodes

[fabien's list]

andys: negation. does this pattern NOT match?
... gets logicians involved
... used for syntax checking a lot

salut, FabienG, ca va?

Jun (... of Oxford): want to traverse an RDF graph with a defined path

scribe: UPDATE

<FabienG> group by and count ;

<FabienG> effective cardinality and counting occurrences of properties and using that count ;

<FabienG> path variables, expressions, path length ;

<FabienG> return expressions in the select ;

<FabienG> direct:: deactivate raisonning for answers ;

<FabienG> distinct on the binding as a set not a list ;

<FabienG> DISPLAY RDF return the RDF/XML instead of SPARQL binding;

<FabienG> MERGE all the results in one graph;

<FabienG> Type operators in filters ;

<FabienG> control the namespaces used to answer ;

<Ivan> merci FabienG

Jun: re service desc: liked darq; it describes what kinds of data are at a source

<FabienG> for more see http://www-sop.inria.fr/edelweiss/software/corese/v2_4_0/manual/index.php#additivefunctionalities

Jun: want to know what anchor holds the data

[negation]

ivan: would like to ask a sparql endpoint to pick from multiple endpoints